For the improvement of the transient driving performance and the thermal efficiency for diesel engines, it is effective to control the fuel injection by model-based control (MBC) on ECU with cycle-by-cycle calculation, and MBC requires six models; gas flow, spray development, mixture formation, combustion, ignition delay, and heat loss. The authors previously developed on-board in-cylinder gas flow and wall heat transfer prediction models to estimate the heat loss. However, the developed gas flow model has an undetermined coefficient called the turbulence intensity coefficient (TIC), which significantly influences the prediction accuracy of the wall heat transfer prediction model. The present study improved the gas flow model and the wall heat transfer prediction model by applying TICs obtained using the PIV and CFD analysis. In-cylinder gas flow in an optical single-cylinder diesel engine was measured by PIV under both motoring and firing conditions, and TICs were calculated and applied to the wall heat transfer prediction model. The heat flux values obtained from the model were compared with those from the experiments using heat flux sensors. It was made clear that the average heat flux using TIC at 60–80 mm plane (stroke of 96.9 mm) showed the least error of -4.7%, suggesting that the use of TIC at planes further away from the cylinder head better predicts the heat flux. For obtaining TICs without experiments, the similar calculation technique of TICs was applied to the CFD analysis, and thus, the relative error of heat flux with TIC estimated at the lower part of cylinder (60–96.9 mm) was -3.8%, which improved the prediction accuracy of the model using the CFD analysis.